[U-Boot] [PATCHv2 1/2] mpc85xx: Initial SP alignment is wrong.
PowerPC mandates SP to be 16 bytes aligned. Furthermore, a stack frame is added, pointing to the reset vector which may in the way when gdb is walking the stack because the reset vector may not accessible depending on emulator settings. Also use a temp register so gdb doesn't pick up intermediate values.
Signed-off-by: Joakim Tjernlund Joakim.Tjernlund@transmode.se ---
v2 - Address Scott Wood's comments arch/powerpc/cpu/mpc85xx/start.S | 16 +++++----------- 1 files changed, 5 insertions(+), 11 deletions(-)
diff --git arch/powerpc/cpu/mpc85xx/start.S arch/powerpc/cpu/mpc85xx/start.S index 8d66cf1..4973682 100644 --- arch/powerpc/cpu/mpc85xx/start.S +++ arch/powerpc/cpu/mpc85xx/start.S @@ -848,18 +848,12 @@ version_string: .globl _start_cont _start_cont: /* Setup the stack in initial RAM,could be L2-as-SRAM or L1 dcache*/ - lis r1,CONFIG_SYS_INIT_RAM_ADDR@h - ori r1,r1,CONFIG_SYS_INIT_SP_OFFSET@l - + lis r3,(CONFIG_SYS_INIT_RAM_ADDR)@h + ori r3,r3,((CONFIG_SYS_INIT_SP_OFFSET-16)&~0xf)@l /* Align to 16 */ li r0,0 - stwu r0,-4(r1) - stwu r0,-4(r1) /* Terminate call chain */ - - stwu r1,-8(r1) /* Save back chain and move SP */ - lis r0,RESET_VECTOR@h /* Address of reset vector */ - ori r0,r0,RESET_VECTOR@l - stwu r1,-8(r1) /* Save back chain and move SP */ - stw r0,+12(r1) /* Save return addr (underflow vect) */ + stw r0,0(r3) /* Terminate Back Chain */ + stw r0,+4(r3) /* NULL return address. */ + mr r1,r3 /* Transfer to SP(r1) */
GET_GOT bl cpu_init_early_f
The code first aligns the SP to 16 then subtract 8, making it 8 bytes aligned. Furthermore the initial stack frame not quite correct either.
Signed-off-by: Joakim Tjernlund Joakim.Tjernlund@transmode.se ---
v2 - Address Scott Wood's comments
arch/powerpc/lib/board.c | 5 ++--- 1 files changed, 2 insertions(+), 3 deletions(-)
diff --git arch/powerpc/lib/board.c arch/powerpc/lib/board.c index d5b75e5..a46942c 100644 --- arch/powerpc/lib/board.c +++ arch/powerpc/lib/board.c @@ -521,9 +521,8 @@ void board_init_f(ulong bootflag) addr_sp -= 16; addr_sp &= ~0xF; s = (ulong *) addr_sp; - *s-- = 0; - *s-- = 0; - addr_sp = (ulong) s; + *s = 0; /* Terminate back chain */ + *++s = 0; /* NULL return address */ debug("Stack Pointer at: %08lx\n", addr_sp);
/*
Ping?
The code first aligns the SP to 16 then subtract 8, making it 8 bytes aligned. Furthermore the initial stack frame not quite correct either.
Signed-off-by: Joakim Tjernlund Joakim.Tjernlund@transmode.se
v2 - Address Scott Wood's comments
arch/powerpc/lib/board.c | 5 ++--- 1 files changed, 2 insertions(+), 3 deletions(-)
diff --git arch/powerpc/lib/board.c arch/powerpc/lib/board.c index d5b75e5..a46942c 100644 --- arch/powerpc/lib/board.c +++ arch/powerpc/lib/board.c @@ -521,9 +521,8 @@ void board_init_f(ulong bootflag) addr_sp -= 16; addr_sp &= ~0xF; s = (ulong *) addr_sp;
- *s-- = 0;
- *s-- = 0;
- addr_sp = (ulong) s;
*s = 0; /* Terminate back chain */
*++s = 0; /* NULL return address */ debug("Stack Pointer at: %08lx\n", addr_sp);
/*
-- 1.7.3.4
U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot
Ping?
PowerPC mandates SP to be 16 bytes aligned. Furthermore, a stack frame is added, pointing to the reset vector which may in the way when gdb is walking the stack because the reset vector may not accessible depending on emulator settings. Also use a temp register so gdb doesn't pick up intermediate values.
Signed-off-by: Joakim Tjernlund Joakim.Tjernlund@transmode.se
v2 - Address Scott Wood's comments arch/powerpc/cpu/mpc85xx/start.S | 16 +++++----------- 1 files changed, 5 insertions(+), 11 deletions(-)
diff --git arch/powerpc/cpu/mpc85xx/start.S arch/powerpc/cpu/mpc85xx/start.S index 8d66cf1..4973682 100644 --- arch/powerpc/cpu/mpc85xx/start.S +++ arch/powerpc/cpu/mpc85xx/start.S @@ -848,18 +848,12 @@ version_string: .globl _start_cont _start_cont: /* Setup the stack in initial RAM,could be L2-as-SRAM or L1 dcache*/
- lis r1,CONFIG_SYS_INIT_RAM_ADDR@h
- ori r1,r1,CONFIG_SYS_INIT_SP_OFFSET@l
- lis r3,(CONFIG_SYS_INIT_RAM_ADDR)@h
- ori r3,r3,((CONFIG_SYS_INIT_SP_OFFSET-16)&~0xf)@l /* Align to 16 */ li r0,0
- stwu r0,-4(r1)
- stwu r0,-4(r1) /* Terminate call chain */
- stwu r1,-8(r1) /* Save back chain and move SP */
- lis r0,RESET_VECTOR@h /* Address of reset vector */
- ori r0,r0,RESET_VECTOR@l
- stwu r1,-8(r1) /* Save back chain and move SP */
- stw r0,+12(r1) /* Save return addr (underflow vect) */
stw r0,0(r3) /* Terminate Back Chain */
stw r0,+4(r3) /* NULL return address. */
mr r1,r3 /* Transfer to SP(r1) */
GET_GOT bl cpu_init_early_f
-- 1.7.3.4
U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot
On 08/14/2012 03:55 PM, Joakim Tjernlund wrote:
Ping?
PowerPC mandates SP to be 16 bytes aligned. Furthermore, a stack frame is added, pointing to the reset vector which may in the way when gdb is walking the stack because the reset vector may not accessible depending on emulator settings. Also use a temp register so gdb doesn't pick up intermediate values.
Signed-off-by: Joakim Tjernlund Joakim.Tjernlund@transmode.se
v2 - Address Scott Wood's comments arch/powerpc/cpu/mpc85xx/start.S | 16 +++++----------- 1 files changed, 5 insertions(+), 11 deletions(-)
Andy Fleming is the 85xx custodian, but you didn't CC him.
-Scott
Scott Wood scottwood@freescale.com wrote on 2012/08/14 23:01:47:
On 08/14/2012 03:55 PM, Joakim Tjernlund wrote:
Ping?
PowerPC mandates SP to be 16 bytes aligned. Furthermore, a stack frame is added, pointing to the reset vector which may in the way when gdb is walking the stack because the reset vector may not accessible depending on emulator settings. Also use a temp register so gdb doesn't pick up intermediate values.
Signed-off-by: Joakim Tjernlund Joakim.Tjernlund@transmode.se
v2 - Address Scott Wood's comments arch/powerpc/cpu/mpc85xx/start.S | 16 +++++----------- 1 files changed, 5 insertions(+), 11 deletions(-)
Andy Fleming is the 85xx custodian, but you didn't CC him.
Oh, didn't notice that Andy was the custodian, there isn't an entry in MAINTAINERS.
Andy, you think you can pick up [PATCHv2 1/2] mpc85xx: Initial SP alignment is wrong.
Jocke
On Jul 23, 2012, at 3:58 PM, Joakim Tjernlund wrote:
PowerPC mandates SP to be 16 bytes aligned. Furthermore, a stack frame is added, pointing to the reset vector which may in the way when gdb is walking the stack because the reset vector may not accessible depending on emulator settings. Also use a temp register so gdb doesn't pick up intermediate values.
Signed-off-by: Joakim Tjernlund Joakim.Tjernlund@transmode.se
v2 - Address Scott Wood's comments arch/powerpc/cpu/mpc85xx/start.S | 16 +++++----------- 1 files changed, 5 insertions(+), 11 deletions(-)
Acked-by: Kumar Gala galak@kernel.crashing.org
- k
Kumar Gala galak@kernel.crashing.org wrote on 2012/08/14 23:28:45:
On Jul 23, 2012, at 3:58 PM, Joakim Tjernlund wrote:
PowerPC mandates SP to be 16 bytes aligned. Furthermore, a stack frame is added, pointing to the reset vector which may in the way when gdb is walking the stack because the reset vector may not accessible depending on emulator settings. Also use a temp register so gdb doesn't pick up intermediate values.
Signed-off-by: Joakim Tjernlund Joakim.Tjernlund@transmode.se
v2 - Address Scott Wood's comments arch/powerpc/cpu/mpc85xx/start.S | 16 +++++----------- 1 files changed, 5 insertions(+), 11 deletions(-)
Acked-by: Kumar Gala galak@kernel.crashing.org
Thanks Kumar
Will you pick up? [PATCHv2 2/2] powerpc: Stack Pointer not properly aligned
It is not 85xx specific but for all PowerPC
Jocke
On 08/15/2012 02:10 AM, Joakim Tjernlund wrote:
Kumar Gala galak@kernel.crashing.org wrote on 2012/08/14 23:28:45:
On Jul 23, 2012, at 3:58 PM, Joakim Tjernlund wrote:
PowerPC mandates SP to be 16 bytes aligned. Furthermore, a stack frame is added, pointing to the reset vector which may in the way when gdb is walking the stack because the reset vector may not accessible depending on emulator settings. Also use a temp register so gdb doesn't pick up intermediate values.
Signed-off-by: Joakim Tjernlund Joakim.Tjernlund@transmode.se
v2 - Address Scott Wood's comments arch/powerpc/cpu/mpc85xx/start.S | 16 +++++----------- 1 files changed, 5 insertions(+), 11 deletions(-)
Acked-by: Kumar Gala galak@kernel.crashing.org
Thanks Kumar
Will you pick up? [PATCHv2 2/2] powerpc: Stack Pointer not properly aligned
It is not 85xx specific but for all PowerPC
I don't think Kumar has a U-Boot tree to pull it into. There's no general PPC custodian. Probably both patches should go via Andy.
-Scott
I will pull them. Possibly tomorrow. I'm currently buried in other work, but Friday is supposed to be my U-Boot Custodian day. :)
Andy On Aug 15, 2012, at 12:13 PM, Scott Wood wrote:
On 08/15/2012 02:10 AM, Joakim Tjernlund wrote:
Kumar Gala galak@kernel.crashing.org wrote on 2012/08/14 23:28:45:
On Jul 23, 2012, at 3:58 PM, Joakim Tjernlund wrote:
PowerPC mandates SP to be 16 bytes aligned. Furthermore, a stack frame is added, pointing to the reset vector which may in the way when gdb is walking the stack because the reset vector may not accessible depending on emulator settings. Also use a temp register so gdb doesn't pick up intermediate values.
Signed-off-by: Joakim Tjernlund Joakim.Tjernlund@transmode.se
v2 - Address Scott Wood's comments arch/powerpc/cpu/mpc85xx/start.S | 16 +++++----------- 1 files changed, 5 insertions(+), 11 deletions(-)
Acked-by: Kumar Gala galak@kernel.crashing.org
Thanks Kumar
Will you pick up? [PATCHv2 2/2] powerpc: Stack Pointer not properly aligned
It is not 85xx specific but for all PowerPC
I don't think Kumar has a U-Boot tree to pull it into. There's no general PPC custodian. Probably both patches should go via Andy.
-Scott
On Mon, Jul 23, 2012 at 3:58 PM, Joakim Tjernlund Joakim.Tjernlund@transmode.se wrote:
PowerPC mandates SP to be 16 bytes aligned. Furthermore, a stack frame is added, pointing to the reset vector which may in the way when gdb is walking the stack because the reset vector may not accessible depending on emulator settings. Also use a temp register so gdb doesn't pick up intermediate values.
Signed-off-by: Joakim Tjernlund Joakim.Tjernlund@transmode.se
v2 - Address Scott Wood's comments arch/powerpc/cpu/mpc85xx/start.S | 16 +++++----------- 1 files changed, 5 insertions(+), 11 deletions(-)
diff --git arch/powerpc/cpu/mpc85xx/start.S arch/powerpc/cpu/mpc85xx/start.S
Why are your patches different from everyone else's? When I try to apply this, I get errors because it can't find "powerpc/cpu/mpc85xx...". git am leaves off the first directory, because the usual practice is to send patches with these filenames:
diff --git a/arch/powerpc/cpu/mpc85xx/start.S b/arch/powerpc/cpu/mpc85xx/start.S
I fixed it by hand, but let's not make that a habit. Is there some setting that I don't have that would make it so I could ignore this?
index 8d66cf1..4973682 100644 --- arch/powerpc/cpu/mpc85xx/start.S +++ arch/powerpc/cpu/mpc85xx/start.S @@ -848,18 +848,12 @@ version_string: .globl _start_cont _start_cont: /* Setup the stack in initial RAM,could be L2-as-SRAM or L1 dcache*/
lis r1,CONFIG_SYS_INIT_RAM_ADDR@hori r1,r1,CONFIG_SYS_INIT_SP_OFFSET@l
lis r3,(CONFIG_SYS_INIT_RAM_ADDR)@hori r3,r3,((CONFIG_SYS_INIT_SP_OFFSET-16)&~0xf)@l /* Align to 16 */ li r0,0
stwu r0,-4(r1)stwu r0,-4(r1) /* Terminate call chain */stwu r1,-8(r1) /* Save back chain and move SP */lis r0,RESET_VECTOR@h /* Address of reset vector */ori r0,r0,RESET_VECTOR@lstwu r1,-8(r1) /* Save back chain and move SP */stw r0,+12(r1) /* Save return addr (underflow vect) */
stw r0,0(r3) /* Terminate Back Chain */stw r0,+4(r3) /* NULL return address. */mr r1,r3 /* Transfer to SP(r1) */ GET_GOT bl cpu_init_early_f-- 1.7.3.4
U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot
Andy Fleming afleming@gmail.com wrote on 2012/08/22 23:08:45:
On Mon, Jul 23, 2012 at 3:58 PM, Joakim Tjernlund Joakim.Tjernlund@transmode.se wrote:
PowerPC mandates SP to be 16 bytes aligned. Furthermore, a stack frame is added, pointing to the reset vector which may in the way when gdb is walking the stack because the reset vector may not accessible depending on emulator settings. Also use a temp register so gdb doesn't pick up intermediate values.
Signed-off-by: Joakim Tjernlund Joakim.Tjernlund@transmode.se
v2 - Address Scott Wood's comments arch/powerpc/cpu/mpc85xx/start.S | 16 +++++----------- 1 files changed, 5 insertions(+), 11 deletions(-)
diff --git arch/powerpc/cpu/mpc85xx/start.S arch/powerpc/cpu/mpc85xx/start.S
Why are your patches different from everyone else's? When I try to apply this, I get errors because it can't find "powerpc/cpu/mpc85xx...". git am leaves off the first directory, because the usual practice is to send patches with these filenames:
Ahh, recently I set (in ny git config): [diff] noprefix = true because I got tired off stripping off that prefix each time I cut and paste file names into emacs and similar.
Seems like git really likes to see that prefix when applying patches. Don't know if git could learn not to complain about missing prefix?
Jocke
On 08/23/2012 02:21 AM, Joakim Tjernlund wrote:
Andy Fleming afleming@gmail.com wrote on 2012/08/22 23:08:45:
On Mon, Jul 23, 2012 at 3:58 PM, Joakim Tjernlund Joakim.Tjernlund@transmode.se wrote:
PowerPC mandates SP to be 16 bytes aligned. Furthermore, a stack frame is added, pointing to the reset vector which may in the way when gdb is walking the stack because the reset vector may not accessible depending on emulator settings. Also use a temp register so gdb doesn't pick up intermediate values.
Signed-off-by: Joakim Tjernlund Joakim.Tjernlund@transmode.se
v2 - Address Scott Wood's comments arch/powerpc/cpu/mpc85xx/start.S | 16 +++++----------- 1 files changed, 5 insertions(+), 11 deletions(-)
diff --git arch/powerpc/cpu/mpc85xx/start.S arch/powerpc/cpu/mpc85xx/start.S
Why are your patches different from everyone else's? When I try to apply this, I get errors because it can't find "powerpc/cpu/mpc85xx...". git am leaves off the first directory, because the usual practice is to send patches with these filenames:
Ahh, recently I set (in ny git config): [diff] noprefix = true because I got tired off stripping off that prefix each time I cut and paste file names into emacs and similar.
Seems like git really likes to see that prefix when applying patches. Don't know if git could learn not to complain about missing prefix?
It's not just git, but also direct use of the patch command when a patch fails to apply cleanly. A user shouldn't have to inspect a patch to determine whether to use -p0 or -p1. -p1 is standard.
How often do you copy and paste filenames out of your own patches?
-Scott
Scott Wood scottwood@freescale.com wrote on 2012/08/23 18:53:14:
On 08/23/2012 02:21 AM, Joakim Tjernlund wrote:
Andy Fleming afleming@gmail.com wrote on 2012/08/22 23:08:45:
On Mon, Jul 23, 2012 at 3:58 PM, Joakim Tjernlund Joakim.Tjernlund@transmode.se wrote:
PowerPC mandates SP to be 16 bytes aligned. Furthermore, a stack frame is added, pointing to the reset vector which may in the way when gdb is walking the stack because the reset vector may not accessible depending on emulator settings. Also use a temp register so gdb doesn't pick up intermediate values.
Signed-off-by: Joakim Tjernlund Joakim.Tjernlund@transmode.se
v2 - Address Scott Wood's comments arch/powerpc/cpu/mpc85xx/start.S | 16 +++++----------- 1 files changed, 5 insertions(+), 11 deletions(-)
diff --git arch/powerpc/cpu/mpc85xx/start.S arch/powerpc/cpu/mpc85xx/start.S
Why are your patches different from everyone else's? When I try to apply this, I get errors because it can't find "powerpc/cpu/mpc85xx...". git am leaves off the first directory, because the usual practice is to send patches with these filenames:
Ahh, recently I set (in ny git config): [diff] noprefix = true because I got tired off stripping off that prefix each time I cut and paste file names into emacs and similar.
Seems like git really likes to see that prefix when applying patches. Don't know if git could learn not to complain about missing prefix?
It's not just git, but also direct use of the patch command when a patch fails to apply cleanly. A user shouldn't have to inspect a patch to determine whether to use -p0 or -p1. -p1 is standard.
Right
How often do you copy and paste filenames out of your own patches?
From patches seldom, from git diff to emacs every now and then.
I guess I will have to change back as I won't remember to switch this feature off before generating patches.
Jocke
participants (6)
-
Andy Fleming -
Andy Fleming -
Joakim Tjernlund -
Joakim Tjernlund
-
Kumar Gala -
Scott Wood